dec :: transcripts :: Jack Gilmore PDP-1 30th Anniversary 1990

DEC — PDP 1 LECTURE
TAPE 1, PAGE 1

JG: My name is Jack Gilmore and I gave the second
lecture of this series [DIGITAL HISTORY] in June. The
first one was given by Bob Everett, covering the
Whirlwind computer. I covered the early graphics work on
the TX-0 while it was at Lincoln Lab and then some
graphics work on the PDP-1 at Itek. Before we begin I'd
like to acknowledge a few people in the audience. First
of all we have one of the fathers, if not the father, of
time sharing, John McCarthy. We also have Marvin Minsky,
father of artificial intelligence, and I believe Ted
Johnson, who did a tremendous amount of work getting
Digital off and running in sales. We are celebrating
today the thirtieth year of the PDP-1, and this is quite
significant. I'd like to welcome members of the press
and the media.

I have two slides that I'd like to pop up and then I'll
turn the meeting over to these youngsters. At the last
lecture series we talked about the TX-0 and the the early
graphics work that was done there, and the advent of the
light pen by Ben Gurley, and the early scope writer
graphics work. Then we pointed out that this was really
the beginning of three other branches, one growing
straight ahead to the more conventional but still very
exciting graphics work that Ivan Sutherland and others
developed on the TX-2.

DEC — PDP 1 LECTURE
TAPE 1, PAGE 2

One other point. There's an individual who isn't with us
who played a very major role in the development of the
work on the TX-0 and on the PDP-1, and that was Ben
Gurley. Ben died a tragic death in '63, but I felt it was
appropriate to remember him as well tonight.

[MOTIONING TO PANEL] You're looking at the "right stuff"
of the early computer world, so dig in and have a good
time.

SL: I wonder if that makes me Tom Wolfe! My name is
Steven Levy. I'm the author of Hackers which had a few
nice words, probably not enough, to say about PDP-1. We
have a terrific panel, people who could tell you all
about the computer which was, as I found out when I
researched the book, an incredibly significant advance
and one which is under appreciated. Rather then my
telling you what I found out from them, let me turn it
over to them, one by one, to tell you about it and then
we'll open things up where you can ask questions and they
can contradict each other and maybe blow some holes in
apocryphal stories or whatever.

Our first panelist is Jack Dennis who, at the time that
PDP-1 was delivered to MIT thirty years ago, had just
joined the faculty as an assistant professor. He's a
professor emeritus, and he is going to talk a bit about

DEC — PDP 1 LECTURE
TAPE 1, PAGE 3

the TX-0 and PDP-1.

JD: The TX-0 arrived at the MIT research laboratory of
electronics in 1959. MIT's Whirlwind computer, which was
the first powerful stored program machine, was
decommissioned in the late 50' s. The main computer
facility then became the IBM 704 installation in building
26 in the MIT computation center. The 704 was basically
a punched card operation. You brought your deck of punch
cards to the door, passed it in to the operators and a
couple of hours or the next day you'd come back and see
what your results were. That kind of operation was very
discouraging to many people, particularly to John
McCarthy here, who wanted to get some really good Al
programs operating and couldn't see how they could be
developed well unless you had more a more suitable way of
interacting with the machine. That was the motivation
behind the development of the time sharing idea and a
little bit in anticipation of that, the TX-0 was run at
MIT as an kind of open shop system, where programmers or
users could come in, have the machine entirely to
themselves for a period time for which they would sign
up. At the beginning what we had to work with from the
Lincoln Laboratory was the hardware, as Jack pointed out
8K of 18-bit memory, and he mentioned that at Lincoln
Laboratory the TX-0 had had 64K of memory. 64K of memory
requires a 16-bit address. With 18-bit words how much

DEC — PDF 1 LECTURE
TAPE 1, PAGE 4

does that leave over for an operation code? It leaves 2
bits. So when TX-0 arrived at MIT, it had a 2 bit
operation code. Talk about reduced instructions set
computersll This was a reduced instructions set
computer! We only had 8K of memory when the machine was
delivered to MIT so we decided to take some advantage of
the three extra bits that were left over, so we
implemented a 5-bit instruction set - slightly less
reduced but also fairly effective for our uses. The way
the machine was used at MIT was as a laboratory tool, a
tool for people in research laboratory electronics to use
to connect to their experiments, to process data from
instrument recordings on magnetic tape, and so on. It
was very much a hands-on operation. To support that
work, some very interesting software developments were
made. We received from Lincoln Laboratory a tape called
UT 3. Utility Tape 3. This was an interactive debugger
but you could only talk to it in terms of octal
addresses, and if you were trying to develop a large
assembly language machine code, it wasn't too convenient
to simply work with octal addresses, so professor Thomas
Stockman and myself developed a program called Flit.
Flexowriter Interrogation Tape. We put that together and
this was a symbolic debugger, it actually could use the
symbolic addresses from the symbol table generated by the
assembler, so another innovative feature it had was it
allowed break point tracing of programs at execution

DEC — PDP 1 LECTURE
TAPE 1, PAGE 5

time. As far as we can tell, this was the first
interactive debugger with those two facilities both
symbolic addresses and a break point tracing of programs.

The other piece of software which we developed for the
TX-0 was a macroassembly language program. This was an
evolution from a tape which we also got from Lincoln
Laboratory. The paper tape we got from Lincoln Laboratory
in this case was for a very primitive assembly language
program, which came as a binary tape with no listing to
it. So the first thing I had to do with it was decompile
it, to find out what it really was doing. That was kind
of tricky because some of the tricks that one could do on
the TX-0 where there's primitive instruction code were
very subtle. [For instance,) after you had completed a
successful comparison, you knew that the number 1 was
left in the accumulator of the machine. A neat trick was
to use that one to index something else that you needed
to index. There are a number of tricks of that sort
which one could use, and this program depended upon it.
I developed a macro assembly language program, a similar
program from that, which was called macro. Eventually
that got translated into a similar program for the PDP-1.

When the PDP-1 arrived as a generous gift to the MIT
electrical engineering department from DEC, John McCarthy
was around with his time sharing so we decided that we

DEC — PDP 1 LECTURE
TAPE 1, PAGE 6

would develop a time sharing system for the PDP-1. That
went into demonstration at the end of 1963, and operation
with users during 1964. This was an unusual form of time
sharing, because we wanted it to support the kind of use
that the TX-0 and the PDP-1 were being put to within the
research laboratory of electronics, which was to work
with various laboratory groups in hands-on computation
within the laboratory. This time sharing system was not
designed for remote use, but designed for shared use of
the machine within the computing facility itself. There
are a couple of interesting things about this time
sharing system. One was that it supported interaction of
time shared programs with user lO devices; that's
something which even modern systems are fairly weak on.
The other thing was that we put in a trap facility into
the PDP-1, such that the debugger could be fully
protected from users programs, so we could run our DDT
debugger developed for the PDP-1 on the basis of the work
on Flit earlier. We developed that so that no matter
what your user program did, it couldn't upset the
debugger in any way. The facilities which we built into
the PDP-1 for that purpose, it seems to me, were taken
over later by DEC in the PDP-11/45. I'd be interested in
knowing a little more about the history of that
development.

I might point out some interesting ideas in this time

DEC — PDP 1 LECTURE
TAPE 1, PAGE 7

sharing system. One was the swapping drum. DEC put
together, with the advice of Ed Fredkin, a drum [that] in
one revolution could exchange the entire content of a
field on the drum, put that in the memory of the computer
and at the same time read the information in the program
on to another field on the drum. We had DEC build two of
these drum systems, one for Ed's machine at Bolt Beranek
and Newman and one for the machine at MIT. At MIT, we
started out with only 4,000 words of memory in our PDP-1,
and how were we going to make a time sharing system with
only 4,000 words of memory? What we did, is wrote a
mere 500 word executive program which sat in the top
piece of this 4,000 words, and did all the work
associated with buffering teletype characters and doing
the primitive scheduling of the machine. Later, we
managed to get enough money together to buy three more
banks of 4,000 words, so we could allow programs to be
either 4,000, 8,000, 12,000, or 16,000 words in length.

A lot of the work in the connection with the TX-0 and the
PDP-1 was done by groups of undergraduate and graduate
students. The undergraduate and other hangers-on became
known as the "hackers." Part of that started because I
have a history, as an undergraduate at MIT, as being
involved with the Tech Model Railroad Club. The Tech
Model Railroad Club was an interesting place because it
had this huge system built with telephone relays that

DEC — PDF 1 LECTURE
TAPE 1, PAGE 8

could manage the operation of the trains around the
layout of the Club's 8 gauge layout. I thought the people
over there might be interested in seeing what the
computer facilities in the electrical engineering
department at the research laboratory in electronics
[were doing], so I invited a whole bunch of them to come
over and see. When they saw what was there, they got
really excited about it, and started making very good use
of these computer facilities. A lot of that is what you
will hear now from my colleagues on the panel. Thank
you.

[APPLAUSE]

SL: Our next panelist is currently a professor of
physics at Boston University, but thirty years ago he was
working at BB&N and laid claim to the title of the best
computer programmer in the world. This is Ed Fredkin.

EF: There was in the Hynes Auditorium, I believe,
something called an Eastern Joint Computer Conference
that I went to. I'd heard that there would be a machine
there from Digital. Digital made modules in those days,
so I went, and there was the prototype PDP-1. It was like
a dream come true. It ran as fast as machines that cost
millions of dollars, but it was priced at $120,000;
actually $80,000 with one thousand words of 18-bit

DEC — PDP 1 LECTURE
TAPE 1, PAGE 9

memory. That's like 2K bytes; that was a lot of memory
in those days. I was watching it, and I ended up right
then and there doing my first bit of debugging on the
machine, because about every three or four minutes the
machine would suddenly stop, and the people there would
look perplexed and start it up again, and say something
like "Well, it's a new machine." So I started snooping
around, and I discovered behind this machine was a large
copper strap with a wire going to the machine. I followed
this strap over to a booth, several booths down, that had
a tape drive with these huge motors that went "punk" like
this, and every time they would push the button to start
the tape drive, the PDP-1 would stop. So asking a few
questions, I discovered that these were the only two
booths that requested a ground wire, so they supplied
them each with a ground wire that just went from one
booth to the other. Unhooking that fixed the problem.
There I ran into Ben Gurley, whom I'd met before — he is
the designer of this machine, it was really a brilliant
concept. I want you to understand how Digital came to the
conclusion of what it was that was to be designed. This
machine was just right. What were the specifications, I
asked? What were you told? What was your charter? I
knew that he was hired by Digital, at some point and was
told to design a computer. He said he had the following
conversation with Ken Olsen and Harlan Anderson who ran
the company. "We would like to hire you to design us a

DEC — PDF 1 LECTURE
TAPE 1, PAGE 10

computer." Ben said, "What kind of a computer?" They
said "Out of inventory." This was a company that
financed itself with very little money, and there was
this stock room, and he was to look through it and see
what was there. The truth is, he had to design about
half the module types that were used in the computer, but
that was the general idea. It was really a brilliant
conception.

I decided that we had to have one of those. I worked at
BBN and I argued and politicked there and we bought the
very first PDP-1 and had an enormous amount of fun with
it. I wanted to mention that John McCarthy and Marvin
Minsky hung around there, and this had a dramatic effect,
I think, on the use of the PDP-1 and computers in
general. All kinds of things were done. Marvin and I
were reminiscing that ever since the PDP-1, we've been
wanting a machine that could do what the PDP-1 did, which
is write some simple code and say, "Here's an XY, put it
up on the screen for me." All displays have gotten so
sophisticated, you can't do that unless you're some kind
gee whiz programmer. Someone should make that possible
again, maybe give Marvin a machine like that and
something new will come out of it. As was mentioned by
Jack, the PDP-1 was one of the machines that pioneered
time-sharing. The other, of course, was the 709 of the
compatible time-sharing system. John McCarthy was at BBN

DEC — PDP 1 LECTURE
TAPE 1, PAGE 11

and one day he explained to me the idea of time sharing.
He had the right idea. We worked at trying to make that
happen and of course Digital helped us. We invented an
interrupt system, and the drum system, and all those
things. They all got built. Those were the days where
you could say, "Hey, I need this new instruction. It's
really terrific." They would go and wire it up and the
machine would have a new instruction.

I'd like to tell you an anecdote which you don't know and
which means that the PDP-1 is, in some sense, still
alive. I started a little company and I located it in
the Mill in Maynard [Massachusetts]. I had a PDP-1,
naturally. One day I brought a friend of mine, John
Cocke from IBM, to look at it. I showed him the machine
and explained to him the architecture. Ben Gurley was
there; he had left Digital and worked at my company. We
had a great time. John Cocke [and Ben were] computer
architects and we went through the machine in great
detail, all the intricacies of its design. It was all
really good stuff. Later on that week, I got a call from
John and he wanted to tell me something funny. He said,
"I was visiting my father. I described this Mill where
your company was, and where Digital was, and my father
said, 'Tell me more about that mill, it sounds
familiar.'" So John kept describing it and then finally
his father said, "I know that mill. The American Woolen

DEC — PDP 1 LECTURE
TAPE 1, PAGE 12

Company used to be there. I used to own it." So, John
Cocke's father used to own the Mill that Digital got
started in. Here's the interesting thing. Many years
later, once late at night, I got a call from John Cocke.
(IBM recently had a John Cocke celebration on a week-end.
They invited people from all over the country and spent
three days doing nothing but heaping honors on John
Cocke. What l discovered is, that there are all these
people, other than myself, who got these late night calls
that lasted two hours or so while John Cocke told all his
interesting ideas. I thought I was the only one until
then.) John said, "Do you remember that machine you
showed me in The mill in Maynard?" I said yes. He said
[he had an idea]. He had worked up a design that was
based on the PDP-1. It had a very simple instruction set
like the PDP-1. PDP-1 had a multiply step and a divide
step instead of an instruction. He'd come up with a
modification of that, that did two bits at a time instead
of one, and handled the sines properly instead of not
handling, and little things like that. He had in mind
that it would have a 16-bit word instead of 18. One
other change he had in mind was the timing cycle; instead
of being 5 microseconds it would be 5 nanoseconds. He
had this all worked out in his mind. He had shown that
this machine could emulate every model of the System 360
in faster than real time, so his idea was that they
should build machines like that, and part of his idea was

DEC — PDP 1 LECTURE
TAPE 1, PAGE 13

to have a very clever compiler. That was different, but
the architecture was the PDP-1. Of course that machine
is alive and well; today, it's called the RS6000. It
went through the 801 and a lot of evolutions, but that
project got started right then and there, and it
absolutely started with the PDP-1 as its idea. To me,
the PDP-1 was one of the great experiences of all time,
and I was always perplexed that they didn't sell
thousands of them. I think they sold 120 or so. One day,
I remember John [McCarthy] and I had this idea that this
was such a great thing, we should convince Digital
somehow, we'll go out and convince other people to buy
these machines. The world took to the idea slowly, but to
those of us who appreciated it, this was a miracle
completely out of left field, there was nothing else like
it anywhere in the world. I think the world owes quite a
debt to Digital for this PDP-1, because it was the germ
that started a tremendous number of things. It was the
first really fun interactive computer. It was fantastic
to sit at the terminal, have the scope there and make it
make music and fancy pictures and started things that
ended up being word processors, and real time control
programs, and all kinds of things that hadn't been done
before because the machines you needed cost millions of
dollars. That was the future that burst on the scene and
captured the imagination of a lot of people. It's a
great story. Thank you.

DEC — PDP 1 LECTURE
TAPE 1, PAGE 14

SL: Our next panelist is currently an electronic
consultant in California, but 30 years ago he was a
research assistant with John McCarthy. He's going to
tell us how he unleashed the curse of the video game upon
the world with the PDP-1. Here's Steve Russell, known as
Slug Russell.

SR: I did see the PDP-1 at the Eastern Joint Computer
Conference, and thought it was interesting but it was
awfully crowded. I didn't have a chance to play with it
at the time. When Digital donated the PDP-1 to MIT I was
down the hall, and came in to see it. I belonged to the
Model Railroad Club. We talked about it a little. I
thought it was a great thing, because it was the first
"appliance" computer I had ever met, and probably one of
the first "appliance" computers. [By that I mean] it had
a switch, you could turn it on, you got a satisfying
clunk and it started working. When you were done, if
there was no one else using it, you turned it off and
left. The other computers of my experience, if they got
turned off, it was a major trauma, and five vacuum tubes
burned out and field service had to come and dance around
it for a while to make it work again. It was also a very
satisfying machine, because you could type a single
character at it and it would type a little message back
you. Granted it was a little cryptic, but it gave you a

DEC — PDP 1 LECTURE
TAPE 1, PAGE 15

great feeling of power, much better than flipping
switches. It had a cathode ray tube, so that you could
draw pictures. I got a real itch to use it and started
thinking about what I could with it. Marvin Minsky had
written a little demonstration program that made
interesting kaleidoscope-like patterns on the machine,
and it was interactive. You could put in different
starting numbers, and it would give you different
patterns. It wasn't terribly exciting, because most of
the time, when you put in new patterns, it didn't work
very well. The play value wasn't really too good. I
wanted to use it for something. We talked it around at
the Model Railroad Club and the Hingham Institute, a dive
on Hingham Street where some of us lived. We eventually
decided that you could probably simulate space ships.
We thought about that for a while, and eventually I got
shamed into writing the code to maneuver a couple of
space ships around the machine in two dimensions. I
fixed it so two people could play. Then there was a great
deal of kibbitzing about the speed at which it ran, and
how there really ought to be gravity [for it] to be
properly realistic. So Dan Edwards wrote some gravity
routines and put them in. Then Pete Samson complained
that the star map wasn't properly realistic, and he put
in a star map. Shag [Graetz] grumbled about the
explosion when he went into hyperspace. It seemed
necessary to escape into hyperspace every now and then.

DEC — PDP 1 LECTURE
TAPE 1, PAGE 16

it helped make the game more playable. So Shag put in
hyperspace and things went on from there. The whole
process when we took a couple of months all told, but it
was great fun, and it created a real problem because it
made it a little hard to use them machine because you had
all these Spacewarl players to kick off. It became
necessary to publish a policy that Spacewarl was
absolutely the lowest priority on the machine.
Fortunately, debugging new versions of Spacewarl was
higher priority. If you think playing video games is
fun, writing video games is even more funl That's how
Spacewarl came to be. It appears that it was the first
two-person interactive game played on a CRT. The people
who made the most money off it were the patent lawyers
who are still discussing it. I think a great deal of
credit goes to John McCarthy and Marvin Minsky and Jack
Dennis, who encouraged everybody around to try things and
were not at all critical, and radiated a lot of
enthusiam, which helped us try all sorts of things that
eventually turned out to be interesting. I think the
most significant thing was what people get in personal
computers today: you sit down, you have something to do,
you poke at the computer, it does something and you find
out about it right away, no muss, no fuss, no bother, no
wait. The PDP-1 really started programmers thinking
about how to do that, and how to do it well. It took
them about fifteen years to figure it out. I think it's

DEC — PDF 1 LECTURE
TAPE 1, PAGE 17

much better now.

SL: Our next speaker, picking up the thread from that, is
currently a free-lance writer but back then he was also
hanging around MIT as a Hingham Institute Fellow,
thinking, of course, about Spacewarl and hyperspace.
This is Shag Graetz. (J.M. Graetz]

SG: I'm glad Slug mentioned the lawyers. Earlier this
week I was down in New York and New Jersey doing some
software detective work, indirectly for the lawyers for a
lawsuit that I call Spacewar East. The people at
Magnavox who have a patent that they are using to insist
they have the authority to license anyone who writes any
kind of interactive game, video game, computer game,
whatever, periodically get sued by other companies who
think, this isn't quite right. In the first place,
Magnavox 's patent only came out in 1967 and of course the
first thing that they adduce as evidence to the contrary
is Spacewar! This has happened about three times over
the last fifteen years. At various times. Slug, Marvin
ninsky, John McKenzie, and others, get called and come
and help out to see what we can do. It's still going on.
The PDP-1 lives both in body and spirit, I guess.
All I want to do is take a slightly different thread or
maybe more than one thread, because it's not just people
who were actively involved in electronics with Digital or

DEC — PDP 1 LECTURE
TAPE 1, PAGE 18

with MIT that got involved with this. That's only one of
the ancestors of Spacewar! I probably have the least
legitimate credentials of all the people here to be
called a hacker. I did go to MIT. The first thread
starts there because one of my first and still one of my
closest friends was Wayne Wiitanen, who was a mathematics
major and one of the few people [during the mid-fifties]
when, at that time at MIT, the only digital computer that
was doing anything, was Whirlwind. But like the TX-0 a
year later, it was also available to people who wanted to
do research. Wayne was one of those who got involved
with Whirlwind. He was, as an undergraduate, a very
early programmer. We became good friends most throughly
activities in the Outing Club and our interest in science
fiction. I flunked out and then I flunked out again.
At the same time, Wayne left because if he had stayed he
would have flunked out. One thing led to another and we
went through our necessary six months in the Army. We
back out, and we lived in a cooperative house for a
little while, and then found our first, genuine, grown-up
apartment, which was on Hingham Street in Cambridgeport,
right by the river. We called it the Hingham Institute
because it made us sound important. It was just a play
thing, really. At that time, I was looking for something
that was better to do than be a research assistant in a
chemistry lab because that's all I really knew. I was
out of work for a while, and it turned out that at

DEC — PDP 1 LECTURE
TAPE 1, PAGE 19

Harvard, where Wayne was working as a programmer at the
Litaur Statistical Lab., they needed a machine operator,
(comma) junior, which is the bottom of the totem pole. I
got that job. It paid about 50 percent more than I could
get as a trained chemist. That started that thread
going. By that time we had become friends with Slug and
we were all avid Outing Clubbers. We went hiking every
chance we got. The other thing we did, every chance we
got, was buy and read, or go down to the movie theatre
and see, some of the out and out worst science fiction
ever written and/or filmed. The novels of Edward E.
Smith, all about the skylark of space and the great
lensmen - 1930 's pulp trash. It is tremendously
exciting. If you pick one up now, even these days, it's
very difficult to put it down if you let yourself go for
just a few paragraphs, but it is bad! The same is true
for all these science fictions movies that came out of
Japan at the time from Toho Studios. We gave it the name
Grade Z science fiction, which is certainly what it was.
The best thing about it was the model work. They built
these beautiful, intricate models of Tokyo, San
Francisco, other places and all these marvelous rubber
monsters, the ones you would be most familiar with would
be Godzilla and Rodan, but there was a whole family of
them. They also had space ship epics with model space
ships, weren't quite as successful somehow. Probably
because we knew more about all that kind of engineering

DEC — PDP 1 LECTURE
TAPE 1, PAGE 20

stuff than we did about monsters and Tokyo. But these
two concepts got fused in our minds, and the first thing
we thought of was, "Well, the obvious thing to do is they
should make movies out of the Skylark of Space." But of
course that would have required the kind of model work
that didn't become feasible really until, probably until
2001. "Star Wars" is probably what the Skylark of Space
would have been. It virtually had the same plot, really.
But, at the same time, we were working, first at Harvard.
Then I left Harvard Statistical Lab and went to work for
another old friend, someone I had met when I was still an
undergraduate at MIT, Professor Jack Dennis, who, at that
time, was in RLE [Research Lab in Electronics] and was
master of the TX-0. He hired me in the summer of 1961 to
write a diagnostic program. This did two things: it gave
me some employment for the summer, and it also set me on
the path I was going to lead as a software writer. I
wrote a diagnostic for the Potter tape unit, that was
just then installed on the TX-0. When that was over, I
went to work for a man that we affectionately called
Zeus. His name is Douglas Ross. At that time, he was
Director of the Electronic Systems Lab, which was in a
nearby building. In the fall of that year, the PDP-1
arrived at MIT. We had been anticipating this for a
number of months, and Wayne and Slug and I would get
togther at Hingham Street and think about, what we could
do to show the machine off. It always revolved around

DEC — PDF 1 LECTURE
TAPE 1, PAGE 21

this business o£ space ships. Between the Toho Studios
and E.E. Smith, it was always on my mind - space ships
moving around on a scope. It didn't take very long for
us to come up with what turned out to be the basic rules
of Spacewar! - two space ships opposing each other,
firing something, rays or torpedoes, moving all around
space and trying to blast the other out of the ether.
We thought it was a good idea, but at the time we didn't
have any immediate way to implement it. Wayne got called
back up into the Army after the Berlin crisis. But as
Slug pointed out, some of the hackers at the Artificial
Intelligence Lab had got involved in the discussions.
They were really quite eager to get going on it. Here
was this machine, it had no software, [it was] just there
- dying to be used. Slug kept making excuse after
excuse, what was he going to do? Couldn't write
anything. Alan Kotok went over to DEC, found a couple of
mathematic subroutines. He came back with them, said,
"Here they are. Write." So starting in January of 1962
Steve sat down, wrote the main control routine and like
Tom Sawyer with the whitewash brush, got everybody else
involved with writing pieces of the program. The first
thing we had were the space ships and the torpedoes and
the acceleration. Then Slug threw in a few random stars.
It was really the display of DDT sitting in upper memory.
[LAUGHTER]. Understand, we did all this is 4K, no more
than 9 kilobytes. Peter Samson didn't like that, so he

DEC — PDP 1 LECTURE
TAPE 1, PAGE 22

wrote the Expensive Planetarium, which displayed the
stars of the central belt around the equator in their
respective magnitudes so that brighter ones looked
brighter. Dan Edwards put in that star in the middle
that had gravity, not particularly Newtonian, but
everybody fell into the star if they didn't move. He
also wrote a routine that compiled the space ship outline
so that there wouldn't be any flicker on the screen. We
were starting to exceed the refresh rate. I was working
on hyperspace. Hyperspace was where we inhaled the
Minskytron. You'll pardon me, that's what we always
called it. Professor Minsky always called it the
"tripost display" but we called it the Minskytron,
because that's what it looked like. And I found a way to
make that display, as the ship went into hyperspace. So
out of these threads - MIT flunk-out, hacker-type
mentalities, science fiction, monster movies, and
something that was a computing equivalent of a Heath kit
— came together in 1962, and Spacewarl came out of that.
It eventually showed up, I think, on every PDP that
Digital ever produced, and is still kicking around all
over the place. It must be on the VAX somewhere. That's
how it all happened. The rest, if you want to blame us
for it, go ahead. We don't mind.

[APPLAUSE] .

DEC — PDP 1 LECTURE
TAPE 1, PAGE 23

SL: Next up is Dave Gross. Currently he's a consulting
software engineer. Thirty years ago he was a student at
MIT, who managed not to be around when the PDP-1 was
delivered, but he returned soon afterwards to find our
birthday machine sitting in the kluge room. Dave?

DG; I was sitting there trying to figure out what I was
going to say and realizing that if this was a meeting of
Hackers Anonymous, I would be up here saying, "I am a
hacker." [LAUGHTER] That's what I was back then. As
long as Shag's admitted it, I was another one who had to
escape from MIT before they flunked me out. During that
escape, the PDP-1 arrived. Before I left, a friend of
mine. Bob Saunders, showed me this little glossy brochure
from some obscure company whose name I can't recall now,
but I think it was Digital. He showed me this brochure
that had two computers that they were going to advertise
for sale: one of them was the PDP-1. I said, "What's
that?" He said, "That's a commercial version of the
TX-0." The other one was the PDP-3, a 36-bit machine
that Digital never actually built. As a matter of fact,
we didn't even have a prototype of it. The PDP-1 was
quite an advance over the TX-0, although it was obviously
a close relative. The TX-0 required some training if you
wanted to turn it on. There wasn't one "on" button and
yes, you did get this satisfying clunk, but then you had
to wait for the power supplies to warm up, because

DEC — PDP 1 LECTURE
TAPE 1, PAGE 24

they were powered by a vacuum tube system. To do this
properly, you had to watch this little timer, and you sat
back and you twiddled your thumbs for a couple of minutes
while it warmed up. Then there was another button to
push that started the clock generator, and if you didn't
do it right, or maybe the button bounced, you might have
gotten two. ..the clock generator was a chain of delay
lines. If you didn't start it up right, you were in
danger of overheating the circuits, I believe, because
the two pulses in the chain, cycling the transistors more
than they were supposed to. You weren't supposed to turn
on that machine unless you were authorized, and I wasn't
authorized. That really pained me. I really suffered. I
knew I knew how to turn on that machine, but heaven help
me if I ever did it, so I didn't. The PDP-1 was quite an
advance. It had those magnetic resonance saturating
transformers, the power supply that lasted all the way
up to the PDPIO generation. Those were the power
supplies with the super heavy transformers where if you
opened the back doors of the later machines that required
more power, the machine tended to tip over. The PDP-1
was where those power supplies started. Another advance
in technology in the PDP-1 was that it had pulse
amplifiers. The TX-0 had a delay line to generate the
pulse chains, and then I believe the pulses would
distribute the logic through plain transformers, which
meant if you had to alter the circuitry somewhere and put

DEC — PDP 1 LECTURE
TAPE 1, PAGE 25

more load on the pulse, you'd be changing not only
amplitude of that pulse, because you're loading the one
transformer, but the change would reflect back through
the transformer into the main delay line, and alter the
amplitudes of pulses other than the one you had. The
TX-0 had this oscilloscope over in the clock generator
pack bay, with little circles on the screen marking where
the tips of the pulses ought to be, and if they weren't
there, the machine was in serious trouble. And heaven
help you if anybody changed the gain on that scope,
because it would really cause a disaster. The PDP-1, on
the other hand, was a relatively reliable machine and it
was really great for us hackers. My first stint in the
computer room was as a witness to Bob Saunders writing
this macro-assembler, it was a really fine product,
probably as good an assembly language as you'll even see
on a PDP-11 class machine. When I got back to the PDP-1,
the work of converting that assembler from the TX-0 to
the PDP-1, had already been completed. But I understand
that Digital was first offering another assembler written
for the one thousand word PDP-1, and it had to run there
with no auxiliary storage. The original PDP-1 had a
bidirectional paper tape reader. I've never seen a
bidirectional paper tape reader. But it was a very fine
mechanism. It had two pinch rollers and two brakes, one
on each side of the reading head. The idea of this
assembler was it was going to punch an intermediate tape.

DEC — PDP 1 LECTURE
TAPE 1, PAGE 26

I believe what you were supposed to do - you took the
source tape and put it in the reader and it would read
pass 1 normally, but then it was supposed to read the
tape in reverse for pass 2.

EF: There was no room in memory to put anything other
than the binary code you were assembling, so it read pass
1, and it turned out that if you read the tape backwards,
every definition arrived just in time for its use.
[LAUGHTER] By some miracle this worked perfectly with no
auxiliary storage needed for definitions.

DG: Needless to say, we were not in love with this
assembler.

EF: I thought it was fine. [LAUGHTER]

DG: The crew consisted of Samson, Saunders, and Kotok
and a few others. They were given a challenge to do the
work in one weekend. It was all night sessions.
That wound up being the successful assembler for the
PDP-1, too. It was a pretty neat job.

SR: It should be noted that the level of human
engineering has still left some of us with the mnemonic
for what to do with the switches when you get done, which
was start to continue, continue to start. [LAUGHTER]

DEC — PDP 1 LECTURE
TAPE 1, PAGE 27

There were the start and continue buttons on the console,
and unfortunately, the meaning of pushing them was
reversed from what you wanted, which was what you had to
remember. We've learned a lot, we've improved the human
engineering of computers a lot. If you wonder what's
been going on between the PDP-1 and the Macintosh, a
great deal of it is learning the hard way - that start
to continue to continue to start, and backward paper tape
- isn't easy to remember and easy to explain, and getting
rid of this nonsense piece by piece.

DG: Both the TX-0 and the PDP-1 were paper tape
machines, but I believe Jack hired this grad student to
do a magnetic tape controller for the TX-0. The grad
student's [was] Gordon Bell, and he actually succeeded in
devising a mag tape. Kotok and I were playing bridge one
night in the TX-0 room, — the TX-0 room was a great
place to play bridge in the hot days of the summer
because it was about the only room at MIT that was air
conditioned for the computer, so it attracted us for more
reasons than just the computer hacking — and we looked
at that magnetic tape and thought there ought to be some
way to save that rewinding that paper tape, and rerunning
it back into the assembler for a pass 2, there ought to
be some way to record the paper tape on the magnetic tape
during pass 1, and play it back on pass 2, and wouldn't
it be neat if we could also put the resulting loadable

DEC — PDP 1 LECTURE
TAPE 1, PAGE 28

image on that magnetic tape, too. Between Kotok and
myself, we came up with a patch for the assembler that
would actually write source code onto the tape during
pass 1 and leave lots of blank spaces between the blocks
so that when it did pass 2 it could also write the
resulting object code and load it from magnetic tape.
That was a kluge that wasn't supposed to be able to be
done, but we managed to get it working and it was a
pretty neat system as such things went. Another hacker
who was there to witness, was Peter Samson's music
program. The TX-0 had a monaural hi-fit unit under the
console and it was attached to probably bit 14 of the
accumulator, or some random bit that was determined by
experiment to make the best sounding noises when you ran
the average program. In fact, you could tell when your
program went into an infinite loop because the noise
would go [MIMICS WHINING NOISE]. Your program was
supposed to make [DIFFERENT NOISE HERE]. So it was a
great tool. You could sit back and play bridtfe and hot
worry about the computer dying on you without your
noticing. Peter Samson thought it would be pretty neat
to use it to make music. He invented this game where he
could compile a tune and play the melody on that speaker.
When the PDP-1 arrived, he realized that that was a
faster machine, with slightly better architecture, and he
could actually code this up to play three voices at the
same time. I remember when I got the first demonstration

DEC ~ PDP 1 LECTURE
TAPE 1, PAGE 29

after I returned to MIT, he had coded up a Bach trio
sonata, which is very good for a music system that could
play three voices. It starts off as a solo for the first
voice, and then it gradually brings in the second and the
third voice until you're playing some fugue and it goes
on complicated variations. It was neat. He played it.
[SINGS] Then came the second voice. I said, "Wow, two
voices." Then the third voice. Then Pete did the hack
of hacks, I think. He had very carefully timed the paper
tape reading loop, so that it was a multiple of the sound
incrementing loop. When the music reached the end of the
buffer, it read the paper tape while continuing to play
the music in tune, which is a pretty neat thing to do on
a machine that had a five microsecond memory cycle. It
would read the tape, and that had any number of people
floored that it could do that, with just minimal buzzing
in the background. When we finally got the successor
machine to the PDP-1, which was supposed to be the PDP-6,
and that machine was faster still, had the bigger
computerware and Pete Samson said, "Ah ha, I can now do a
six voice machine." Indeed, we designed something called
the MK-6, which stood for the Music Kluge Six, which
could connect your PDP-6 to a speaker and indeed he did
have a six part music played on that machine.

Another thing developed on the PDP-1 that I think is
significant is a product of a fellow named Dan Murphy

DEC — PDF 1 LECTURE
TAPE 1, PAGE 30

here in the audience, and Mr. TECO. I remember Dan
debugging TECO, and thinking what a crazy thing to do.
The right way to edit a paper tape is to use those
Flexowriters. I'm going to admit it. I'm an old stick in
the mud, and I still like the coding in machine language
where a machine only had a two bit op code; I think
that's what real programmers ought to do.

[APPLAUSE]

SL; Before we throw it open, I think I would be a shame
not to hear, if they'd be willing, from some of our
guests out here in the audience. I think particularly
Drs. Minsky, McCarthy and Richard Greenblatt. Would you
like to say a couple words? Marvin Minsky.

[END OF TAPE]

DEC — PDF 1 LECTURE
TAPE 2, PAGE 1

MM: You were talking about flunking out [of MIT] and most
of the people who made the biggest innovations in this
field did in fact flunk out. In fact, none of them
flunked out they all disappeared right? I've never heard
of anyone [at MIT] flunking out — it happens, but the
point is that it wasn't until 1970 that the professors
could be said to know more then the students in computer
science. They had a kind of artificial view, and I think
this may be a real phenomenon. It certainly is why so
many of these discoveries came from below; the image that
Jack Dennis was head of RLE is a perfectly reasonable
view from the bottom.

[LAUGHTER]

He's all the way up there at at assistant professor rank.
Everything everyone said brought dozens of images [to
mind], like going over to Hefron's and getting the right
switches to play Spacewarl Eli's is still there but it's
no use anymore because it only sells parts from
computers. [LAUGHTER] There was a great junkyard which
had every possible kind of surplus but it had surplus you
could use, because if you're doing anything with
computers you obviously don't want computer parts. One
of the ironies there. There was a nice moment of banning
Spacewar! just for a little while, of course. Many years
later, some town, Braintree, or somewhere, banned arcade

DEC — PDF 1 LECTURE
TAPE 2, PAGE 2

games and I remember thinking "Oh, they thought of that,
too." [LAUGHTER]

One of the great mysteries to me of this whole period was
the disappearance of the graphic display. DEC started it
pretty much. I was just mentioning to McCarthy it's
really hard to believe the 704 was before the PDP-1 and I
have this pretty well confused in my mind, but of course,
the 704 was a bigger, more powerful computer but you
couldn't program it much, and I don't think anybody's
even mentioned Steve Russell's role in the programming of
the first versions of LISP. He played an immensely
important role in that. But that was mostly on the
'other computer'. I remember when we got the PDP-6 one
day, about '64. It turned up, and I said, "Well now we've
got our own real computer. How are we going to do our AI
research on it with no LISP?" At the end of the next
weekend there was. So a great deal of the success that
I'm credited with, and John is in the AI lab, came from
the fact that all this research was BP {Before
Programming] and the same people who did these hacker
exploits also did an immense amount of serious scientific
work, mostly by figuring out what was going to be needed
next year in AI. So when I said "Let's have a LISP," and
they said "Well, it'll take at least three or four days."
[LAUGHTER] There was this decay of graphics; the PDP-1
had this thing and you could, say, plot XY. The only

DEC ~ PDP 1 LECTURE
TAPE 2, PAGE 3

other machine that I've been able to plot XY on is some
little Casio pocket computer which has a plot XY and it's
Basic, but it's almost impossible to plot XY in any real
computer. So in order to do graphics research, you have
to get an awful lot of equipment set up and it'll go to
an awful lot of trouble. Later I built one myself. I
built a little computer for working with for children to
be able to run LOGO programs. Unfortunately, it cost
more then anybody thought, and schools couldn't afford it
but it had a vector plotting scheme and that was great
fun. Shortly after that I saw one in a little computer
game named VECTREX which I guess has bit the dust. As
soon as raster displays came out, graphics went down hill
and took many years to recover. There are lots of other
wonderful stories but there are too many... it's getting
too late.

[APPLAUSE]

SL: Mr. McCarthy, would you like to say a couple of
words here? John McCarthy.

MCCARTHY: I don't have very much to add, to what has
already been said. I would like to compare a little bit
the psychology of that time with the psychology of
computer use today. Maybe I am guessing here, because I
don't really know too much about either one, but consider

DEC — PDP 1 LECTURE
TAPE 2, PAGE 4

that the PDP-1 as people said started with a IK memory
and went up to a 4K memory and then somewhat more, and
now people are talking that one megabyte is rather small;
Harvard is letting me use a machine with eight megabytes
just as a terminal and so forth. Now, I have not
actually learned to write programs 8,000 times as fast as
I did many years ago, in fact I probably write somewhat
slower then I did a number of years ago. While people are
somewhat faster, they aren't certainly 8,000 times as
fast as the people many years ago. The result is, of
course, that people have to program at, so to speak, more
like an executive level, where they don't really know
what all these pieces that they are ordering around are.
Like the boss of a company who doesn't fully understand
what his henchmen are doing, or thinking, or going but he
gives them some orders and hopes that the right thing
will happen. Now it seems to me that a programmer is a
kind of executive, putting together these parts. What is
always amazing me, is people are saying "Well, three
megabytes is not really enough. You need more," and of
course the answer is if they really knew what was going
on, and were in a position to change it, even three
megabytes for many of these things would be plenty. I
had something else I was going to say but I forgot what
it was so I think I'll stop.

[APPLAUSE]

DEC ~ PDP 1 LECTURE
TAPE 2, PAGE 5

SL: Richard Greenblatt, one of the canonical hackers
from that period, will say a couple of things then I will
allow time for a couple of questions.

GREENBLAT: I was really a late comer to the PDP-1 scene.
I arrived as a undergraduate, and never was an official
user, but [here's] what I did. The sign up list went up
on Friday morning at eight o'clock, and within an hour
the entire week for the following week was signed up 24
hours a day. Users coming in and signing up. X would
come along and look at the signup list and see who had
signed up, and try to figure out who was usually late for
their computer time, and I would then wait [until] the
appointed hour and somebody would be a few minutes late
showing up for their computer time and I would jump on
the machine, and play around, and do my thing and then
maybe, maybe, they wouldn't show up for their time at all
or they would come in fifteen minutes late. That was
computer access in those days. Later I was involved in a
number of the things that have been alluded to.

[APPLAUSE]

MCCARTHY: The Sale system at Stanford will — which is a
was originallly a PDP-6 and went through various versions
of PDP-10 and is a KL 10 and is of course thoroughly

DEC — PDP 1 LECTURE
TAPE 2, PAGE 6

obsolete but nevertheless still working, at least the new
service works and so forth, will if it lasts, be turned
off on June 8, of next year which will be its 25th
anniversary. I've been thinking about how it should
celebrate its demise, and it should send people a
message, I think. If anyone wants to be sent the final
message of the of the Sale computer just before it's
turned off, then you should communicate. Now having sort
of thought of that from an administrative point of view
it's not instantly clear to me as to where you should
send this E mail. Well, my poor secretary, I'll make her
do it. [LAUGHTER] If you want to be sent E mail when
Sale is turned off, send E mail to MPS at CS.Stanford.EDU
and you'll get the final message out of the PDP-10,
provided it doesn't crash irrevocably in the meantime
because if it crashes in the meantime it's sort of been
agreed that no heroic measures will be undertaken. It
has asked for that. It has this living will.

SL: Would Ted Johnson like to come up here and tell us
about selling the machine as DEC'S VP of sales for twenty
something years.

TJ: I hope I don't contaminate this hacker's delight
by throwing in a commercial dimension, but I think I can
add a few facts here and a different perspective. As Ed
Fredkin pointed out, before the PDP-1, Digital was

DEC — PDF 1 LECTURE
TAPE 2, PAGE 7

entirely a modules company. You're looking at the sales
force at the time that the PDP-1 was invented. I was the
only person in the field. We had no field, only a person
in sales officially, that is. Everybody was really
selling, but I was the only sales engineer, and we had no
field officers. I was in Maynard. One exception to the
module business is that we also sold memory testers.
Even though I was the only sales person out selling
modules, they took me out of the field for about two
weeks to design the first memory tester with the logic
when after talking with RCA the company decided to build
their own system. I didn't realize they were laying the
groundwork for getting in the computer business, at all.
Our total focus was the module business, and I really
didn't know all the cards that Ken was planning to play,
although I had seen the original business plan. I
probably should have known better, but we were really
focused. In April of '59, I happened to be in the office
with Ken and Harlan Anderson when a request for a
quotation came in. It was from the US Naval Ordinance
Test Station in Pasadena, and there was a request for a
36-bit machine, five mega cycles, fit the module line
perfectly, and Ken turned to me and he said, "This is
just what I wanted to build. Go sell a computer." What's
a computer? I didn't really know. After thinking about
it, and talking to Ben Gurley who joined in June, the
decision was made that if we were going into the computer

DEC ~ PDF 1 LECTURE
TAPE 2, PAGE 8

business, the 36-bit machine would take awhile not only
to build but also to close the sale with the Navy, why
don't we start small and build an 18-bit machine? By the
way, we should probably build one in the middle
someplace. I've never known whether that was supposed to
be 24- or 27- bits but that's the reason why there was a
gap. There was the PDP-1, the PDP -2, and the PDP-3. I
believe the PDP-3 actually was built, by the way. We had
a line of 10 megacycle modules and many, many years later
I was showing that machine, over in Waltham, someplace in
a basement by some drug company.

EF: They ordered one and DEC declined to build it.
Actually two were ordered. AFCRL ordered one. They were
delivered two PDP-1' s, with some explanation that that

was the [LAUGHTER] The other one was a company, an

architectural firm, who really wanted it so they said,
"Do you have any pieces of paper with any kind of
design?" They took the design, bought the modules and
built a machine, and made it work.

TJ: Ben Gurley joined in June and began designing and
laying out the PDP-1. We introduced a working machine in
November, so that was a tremendous time to market, and
from our perspective selling modules, it was just a
testimony to what you could really do with using our
standard modules. The one competitor that we had the

DEC — PDF 1 LECTURE
TAPE 2, PAGE 9

first year that I recall was the CDC 160-A. I have a
slightly interesting story there, too. I think most of
you might know that Digital and CDC started off in the
same month, same year - September of 1957 - both going
off in the scientific engineering area but totally
different strategies. My module selling was very
difficult; it was difficult to find a customer for
modules in those days, and I remember calling up CDC at
one point and visiting them in Minnesota and trying to
sell them on converting and using our modules to build
their computer. Of course, they were much too far along
and it was a silly idea, but I was naive and an eager
salesperson. I did see in the side when they were
building the 1604, and the 160A was the prototype, the
test machine, that they built, and it was sort of a side
line for them to go out and try to sell this machine. I
think they were the first ones to make a deal, an OEM
contract, for small machines. From then on the brass
largely focused on the real selling of PDP-l's for a
couple of years, and the sales force — by now I was on
the west coast — continued to largely focus on selling
modules. We did sell 49 PDP-l's. Sixteen of those were
sold to ITT, who used them for message switching
communications applications. Nick Mazzarese was brought
in to be the account manager for ITT. That was really
the start of the computer OEM business, and of course,
later on in 1965 when we were into the product line

DEC — PDP 1 LECTURE
TAPE 2, PAGE 10

organization, Nick was put in charge of all the small
computers. I think that's just some added perspective.
Thank you.

[APPLAUSE]

EF: I'd just like to make a comment. When DEC got
started, one ground rule they had was no software. That
was for other people: You want this machine, you guys
write the software. About selling the computers I
remember that one day we had a PDP-1 at BBN and Harlan
Anderson, who was a partner with Ken Olsen running the
company, called me and asked me to go with him on a sales
call for the PDP-1. I agreed. We went off to Ohio State
University, and talked up the PDP-1, and it was clear
that they weren't going to give an order right then and
there. As we were coming back on the plane, Harlan
Anderson grumbled to me, "That's the last time we'll ever
do that." I said, "That's the last time you'll ever do
what?" He said, "Call on someone to try and sell them a
computer. If someone wants one of these machines they
got to come here and buy it..."

[LAUGHTER]

SL: Before I throw it open for some questions, I actually
have a question. There was one story which I determined

DEC — PDF 1 LECTURE
TAPE 2, PAGE 11

at least to be at least partially apocryphal. It
concerned a little wire that some of the hackers hooked
up between the TX-0 and the PDP-1 to play a prank on the
professors - Minsky and McCarthy. [It involved] hooking
up this computer, giving it the software to play chess,
and one was in one room and the other was in the other
room and they were actually playing against each other,
thinking they were playing against the computer until one
of them, I think as the story goes, McCarthy noticed that
the moves were being really put on one letter at a time
as a person would type it. He walked in the other room
and discovered it. I think I eliminated Marvin Minsky

from the story and then when I asked him about it *.

Do you remember that at all?

MM: The part about me is correct. The conjecture was that
I was playing against some human but it just seems to me
I wasn't actually told against whom I was playing so I
just assumed that it was one of the hackers. Is that
true or not?

SL: I actually attributed the story to Samson, the way
he told it, after I eliminated Minsky. He insisted the
idea was that you were playing against the computer not
another human

MM: Yes it was certainly billed that way.

DEC — PDF 1 LECTURE
TAPE 2, PAGE 12

SL: Do we have any questions from out here for any of
our panelists?

AUDIENCE: Steve, tell me why weren't the torpedoes in
Spacewarl never affected by gravity?

SR: Because it took too much time to calculate gravity
for the torpedoes. An example of an important principle
of PDP-1 programming, which was the dominating principle
of PDP-8 programming, was that you compromise the problem
until it fits on the machine you've got.

AUDIENCE: DEC sells a game book with all the original
games that were on all the original PDPs and I think
Spacewarl is published in that book

SR: That's the Dave Ahl book isn't it? Basic Computer
Games ?

AUDIENCE: One of the original books and I'm assuming
that Spacewarl originally was not written .

SR: I used the most sophisticated language available at
the time I started, which was Macro.

SGi

Spacewarl was written in a language that you will

DEC — PDP 1 LECTURE
TAPE 2, PAGE 13

find in this particular book. There are two separate
listings of Spacewarl The original final version and a
version that was developed at MIT over the ensuing summer
months and that came out in September. If you want to
see what the kind of programming that we did, you're
welcome to have a look at this afterwards. I'll open it
to a few pages, and you'll get some idea of what went
through our minds.

SR: I would like to point out with some unjustified
pride that I was quizzed by the legal consultants a year
or so ago about what Spacewar! did and I had commented it
just barely well enough so that I could look at what was
puzzling them and say, "Oh, yes that does," and explain
it correctly.

[ LAUGHTER 1

DAN MURPHY: Do any of you know the origin of TECO?

SR: It's all his fault! The guy with the yellow shirt.
He did it. He's the last guy to design the whole thing.

DM: Alright. I plead guilty. TECO was cooked up on the
very same PDP-1 that we've all been talking about, on the
second floor of Building 26. [DIGITAL MILL IN MAYNARD]

DEC — PDF 1 LECTURE
TAPE 2, PAGE 14

SL: TECO, for the benefit of those who might not know
was,

DM: It was a Text Editing program, one of the early ones
that you could use interactively, and it had a couple of
things that were advances over what seem to be kicking
around at the time. When I first arrived at the PDP-1
there was a program out called the Expensive Typewriter,
which you could in fact use to change your program using

the computer and you the tape. It had

two problems. One, you had to use it on the computer and
secondly, you could only change the whole line of your
program at the time, so if one letter was wrong you had
to retype the whole thing. My motivation for writing
TECO was to change both of those aspects, to let you be
able to change one character at a time, but also to save
what changes you wanted to make, off line, using the
flexowriters in the next room. So when you got your
precious little shot of machine time, you could go there,
take your program, make the changes, and have a new tape
punched out all in a very short order. It turns out, of
course, that TECO was almost never used in that mode. We
very quickly added a switch where you could enter the
command while you were on the computer, and of course
that's what everybody then did with it for all the years
later that it was used. I want to tell one other story
about TECO. Ever since your book came out [ Hackers;

DEC — PDP 1 LECTURE
TAPE 2, PAGE 15

Heroes of the Industrial Revolution ] I just wanted to
point that some of the hackers in MIT in that era, not
all of them, came from the Model Railroad Club. There
were a few of us who came from the MIT radio station. I
think this is significant because one's background tends
to have some influence on the reasons that one conducts a
certain task. With TECO I had a version of it that did
just the very basic kinds of editing things, replacing
characters and finding stuff and so forth and the next
level of improvements to it came about this way. At the
radio station we had a guy who was an interesting talent,
and this story's a little embarrassing, but here's what
it was. He was able to read copy like news reports and
so forth in a simulated Chinese accent by virtue of
merely exchanging all the L's and R's of the text and he
could do this in real time. Like saying. Digital,
"Digitar histoly recture series." So sitting around [a
Chinese restaurant] one night we decided it would be
really nice if a computer could change all the L's and
R's and keep checks so that anybody could read in this
pseudo-Chinese manner, so you go through and you look for
the L's and you change it to something else. From that
came the loop capabilities, the conditional capabilities,
and several other things at TECO that turned out to make
it rather useful.

SL: So we all owe search and replace in our

DEC — PDF 1 LECTURE
TAPE 2, PAGE 16

wordprocessors to that hack? is that it?

MM: In fact I did both by programming in the sixties in
TECO, and I still have this little bit of code which I
think is the shortest description of the universal Turing
machine ever written, about four little lines of TECO.
But the last time I tried it it didn't work. I wonder if
you have a up to date version of TECO running on
anything.

DM: You may be interested to know, I had nothing to do
with it, but some of the people in the VMS group had in
fact produced a native version of TECO for VAX/VMS in
the past few years. I don't know if it had enough
[INAUDIBLE]

SR: For those of you who don't know TECO, you should
understand that those who do, have some form of love/hate
relationships with it. It is very powerful, and we've
all used it to do something that we couldn't do any other
way. But it's also powerfully mysterious when you start
trying to do things that are complicated, and we've all
been done-in by it more times then we can remember.

MM: It's good because it makes APL look so simple.

[ LAUGHTER ]

DEC — PDP 1 LECTURE
TAPE 2, PAGE 17

EF: I don't know what the ultimate TECO hack is, but I
consider the macro written by Gosper that found the first
glider gun by searching through every kind of
configuration for the game of life. This was discovered
by a TECO macro.

DG: For many years at DEC, TECO was an important CAD tool
too; we couldn't have done the PDP-10 without it.

SR: And vital to Digital's engineering data processing
because Dick Best's list was maintained in TECO.

AUDIENCE: Many years ago at DECUS, I think they were
celebrating the tenth anniversary, there was a big trivia
contest. One of the questions was 'what does TECO stand
for?' What's the real answer?

DM: I do get asked that question occasionally and in fact
it was 'Tape' because the only thing you could do on

PDP-1 was take the tape or take the and punch

out a new one and so in fact the T for tape wound up in
several places. DDT - DEC debugging tape.

SR: I guess we should also pay homage to another family
of PDP-1 programs. We mentioned Expensive Typewriter, it
was also Expensive Desk Calculator, and Expensive

DEC — PDP 1 LECTURE
TAPE 2, PAGE 18

Planetarium, and at Stanford when we had a PDP-1 we also
briefly had Expensive Tape Recorder and Expensive Mirror.
The idea was these things did the functions they said
they did but they did them using a $120,000 worth of
equipment which made them spectacularly more expensive
then using the real item.

AUDIENCE: Was did Expensive Mirror do?

SR: It was a program that read the TV camera with
excruciating slowness and displayed it on the CRT.
[LAUGHTER] SMOP : meaning, Small Manner of Programming.

SL: There was one more question out there okay?

AUDIENCE: I was just going to mention that there was
another interesting piece of equipment on the other PDP-1
1 at MIT at the nuclear science lab. That was about a 58K
machine and DEC hadn't figured out all the problems with
shipping things. What they did for adding more memory
was bolt in an expanded cabinet in the back. Then you
wired across the cabinet frames. When it got to twenty
feet long it was 58K of core, it couldn't fit in the
elevator so they actually had to hoist it up and enter
through a window in order to deliver it. [INAUDIBLE]. It
was a piece of equipment on it the IBM typewriter
that

DEC — PDF 1 LECTURE
TAPE 2, PAGE 19

MAN: The Soroban Compu-typer

MAN: We had an IBM ball typewriter. It's kind of a
interesting characteristic. The program we had we needed
to call the operator to input things - we were doing film
scanning for high energy physics. The operators would be
out of the room, so what they would do to call them when
when they needed input. They would type out a message
saying please input and wait a bit. If nothing happened
it would switch the ribbon from black to red and then
back again. That was the command you'd be given
[INAUDIBLE] If nothing happened after a few seconds of
doing this, and the operator didn't arrive, it would then
shift the whole keyboard. The whole carriage would go
grumf, grumf. You'd walk down the hall and here's this
typewriter going like that... it was our signal for the
operator.

SR: That reminds me of one of the things that we didn't
mention about PDP-1 which I was noticing as I went over
to the other building, is that they all came with a piece
of do it yourself I/O in it. One of the things that you
could do merely by purchasing a few extra modules was
cause whatever you had in the room to get connected to a
computer, and almost everybody used that for getting
Spacewarl controls in, but it was also done for a lot of

DEC — PDP 1 LECTURE
TAPE 2, PAGE 20

lab equipment. Nobody else had ever even thought of
telling you how to get your switches into the computer or
get your signals out. It's just you plugged it into the
official printer and that was all you needed to do.

EF: When the first production PDP-1 was delivered to BBN
it had this kind of 10 capability, and we had this
scheduled event, and Ken Olsen was there and all kinds of
important people. We had a ribbon cutting ceremony
planned, except the computer actually did something that
pulled that paper cutter down and it chopped its own
ribbon once.

The greatest thing about the PDP-1 was the fact that when
you had an idea — this was the first time in the history
of the world and this idea involved either programming or
hardware ideas — you could implement them in a few days
and all kinds of very important things that used to be
projects that took forever just a bunch of guys would go
and make something happen. That illustrated that it could
be done, and they'd do it by staying up a few nights in a
row and have it done in a few days. It was just
fantastic.

MCCARTHY: I want to mention something else that perhaps
I should have mentioned before, and that was the PDP-1
was a machine that had a five microsecond cycle, yet it

DEC — PDP 1 LECTURE
TAPE 2, PAGE 21

it was possible to build time sharing systems on it that
served a fair number of users better than many systems do
today, in terms of promptness. Swapping drum, of course,
was the key to it, in that in one drum revolution which
was thirty milliseconds without latency - another piece
of its elegance — it could swap one 4K user out and
another 4K user in. So when we had it at Stanford, it
could really give reasonable service on this very slow
machine to twelve users. I remember when we started the
LOTS, the Low Overhead Timesharing at Stanford, which was
in '75 and we used DEC-20, I had made calculations back
in the early days which said that PDP-1 ought to be able
to handle quite a lot of users if all they're doing is
editing because here's how long it takes for an interrupt
and here's how long it takes to put the character in the
buffer. Then the question arose, well. Why is it that
one DEC-20 is not enough to handle these students? Why
is it so slow? I put a guy to working on it and he found
out that if you were using the ancient editor then as it
was written at Stanford a long time ago, then a typical
editor user used one 250th of the DEC-20, which was
pretty bad from the point of view of what ought to have
been possible, but was plenty good from the point of view
of our requirements. However, if he was using Emax, he
was using l/70th of the machine, which made him an
average user, that is we were trying to serve 70 people
on the machine. One of the other consequences of this

DEC — PDP 1 LECTURE
TAPE 2, PAGE 22

enormous increase in memory is that nobody knows today
where the computer time is going.

SR: It goes into running all that code that fits in all
the extra memory.

AUDIENCE: We've got these machines with all this memory
so how many virtual PDP-l's are there? You guys love
using PDP-l's they're color scopes, there are black and
white scopes. Are there lots of emulators running around
for this great machine?

[ LAUGHTER ]

SR: I don't think anyone claimed it ultimately great for
now. It seemed great at the time. It was a paper tape
machine and at Stanford the first time sharing system we
built there we faithfully simulated paper tape on the
drum. We then tried to use it for some undergraduate
courses, and discovered that paper tape was pretty
cumbersome to explain. It was a forty-six step process
to get your program assembled when you wrote it all down
and when you faithfully simulated it, the fact that you
didn't have physical tape didn't make it a bit easier to
understand at all. It really is nice to have a proper
file system.

DEC — PDF 1 LECTURE
TAPE 2, PAGE 23

SL: On that note let's thank our panel and our guests
for a terrific afternoon. And thank the PDP-1 for kicking
off a year of interactive computing.

[END OF SIDE 2]